Skip to main content

Table 2 Model performance

From: Evaluation of word embedding models to extract and predict surgical data in breast cancer

Metric

File

Method

Score

Similarity

WS353

Pearson correlation

Spearman correlation

OOV ratio

0.34

0.37

51.71

Similarity

UMNSRS449

Pearson correlation

Spearman correlation

OOV ratio

0.62

0.69

91.31

Similarity

SimLex999

Pearson correlation

Spearman correlation

OOV ratio

0.15

0.13

52.25

Analogy

Google Analogy

Accuracy

0.23

  1. Metrics are listed, obtained using tests in the generic domain (WS353 and SimLex999) and clinical domain (UMNSRS449)